Microbial Production of Xylitol

APPuED MICROBIOLOGY, Dec. 1969, p. 1031-1035 Vol. 18, No. 6 Copyright ( 1969 American Society for Microbiology Printed in" U.S.A. Microbial Production of Xylitol from Glucose' HIROSHI ONISHI AND TOSHIYUKI SUZUKI Noda Institute for Scientific Research, Noda-shi, Chiba-ken, Japan Received for publication 22 July 1969 A microbiological method is described for the production of xylitol, which is used as a sugar substitute for diabetics. A sequential fermentation process yielded 9.0 g of xylitol from 77.5 g of glucose via D-arabitol and D-xylulose. Candida guilliermondii var. soya (ATCC 20216) consumed 5.1 g of D-xylulose and produced 2.8 g of xylitol per 100 ml. Pentitol production from D-xylulose by yeasts was divided into three types: I, yeast-produced xylitol; II, yeast-produced D-arabitol; and III, yeast-produced xylitol and D-arabitol. D-Xylulose, but not glucose, was dissimilated to xylitol by yeasts under aerobic conditions. A large variety of polyalcohols are produced in good yields by aerobic dissimilation of various pentoses and hexoses by yeasts (10-15, 20-22). No specific commercial outlet for these polyalcohols has been found. Recently, however, it has been shown that xylitol may be used as a sugar substitute for diabetics (5, 7), and xylitol is now commercially available for this purpose in Japan. Xylitol is readily produced chemically (6) or biologically (13) by the reduction of D-xylose, but the process is expensive because of the high cost of the substrate (D-xylose). It has been well demonstrated that glycerol, erythritol, D-arabitol, and D-mannitol are produced by aerobic dissimilation of glucose by yeasts (10, 11, 15, 20-22), but xylitol has not so far been found to be a fermentation product of glucose. We examined microbiological conversion of glucose to xylitol. Although we failed to obtain xylitol from glucose by one fermentation step, a microbiological method was developed which converted glucose to xylitol via D-arabitol and D-xylulose, in which three sequential steps are involved. MATERIALS AND METHODS Yeast strains examined. A total of 128 yeast strains were screened for their ability to produce xylitol from glucose and D-xylulose. Most of them are from the collection of our institute. They were: 2 strains of Saccharomyces acidifaciens, 2 strains of S. acidifaciens var. halomembranis, S. cerevisiae, S. fermentati, S. fragilis, S. logos, S. mellis, 44 strains of S. I A preliminary report on this topic was presented at the annual meeting of the Agricultural and Chemical Society of Japan, 1-4 April 1969. rouxii, 6 strains of S. rouxii var. halomembranis, 16 strains of Debaryomyces hansenii, 3 strains of Pichia farinosa, P. hyalospora, 3 strains of P. membranaefaciens, P. polymorpha, P. pseudopolymorpha, P. quercuwn, P. rhodanensis, P. robertsdi, P. scolyti, P. wickerhamli, Hansenulda anomala, H. anomala var. productiva, H. saturnus, H. schneggii, H. suaveolens, H. subpelliculosa, 4 strains of Hansenula species isolated from miso paste, Endomycopsis chodatii, Candida albicans, 2 strains of C. guilliermondii, C. guilliermondii var. soya, C. melibiosii, 2 strains of C. parapsilosis, C. parapsilosis var. intermedia, C. polymorpha, 5 strains of C. tropicalis, C. utilis, 3 strains of Candida species, 2 strains of Torulopsis famata, Torulopsis halophila, Torulopsis sake, Torulopsis versatilis, Trichosporon cutanewn, Trigonopsis variabilis, Cryptococcus neoformans, Cryptococcus laurentil, and Rhodotorula rosa. Media. The composition of the standard medium used for screening of xylitol production from glucose (medium A) or D-xylulose (medium B) was as follows. Medium A: 10% glucose, 0.1% KH2PO4, 0.05% MgSO4-7H20, 0.01% CaCl2-2{HaO, 0.01% NaCl, 0.4% vitamin-free Casamino Acids (Difco), 0.1% Yeast Extract (Difco), ph 5.0. Medium B: 5% D-xylulose was used instead of 10% glucose in medium A and added aseptically after sterilizing separately at 110 C for 5 min. Changes in the composition of the medium are noted. Fermentation conditions. To screen for xylitol production from glucose, one loopful of yeast from a 5-day koji-agar slant culture was inoculated into large test tubes, each containing 8 ml of sterilized medium A. The tubes were shaken on a reciprocal shaker at 300 rev/min with a stroke of 2 cm at 30 C for 2 to 5 days. The reduction of D-xylulose to xylitol and some factors affecting the fermentation were also examined under the same conditions except that 10 ml of medium B was used. Fermentation temperature of the latter was 30 C unless stated otherwise. 1031

1032 ONISHI AND SUZUKI APPL. MICROBIOL. Analytical methods. After removal of yeast cells by filtration or centrifugation, the clear fermented broth was analyzed for sugar and polyalcohol by the method of Neish (9) and for D-xylulose by the method of Dische and Borenfreund (2) with crystalline D-xylulose p-bromophenylhydrazone as a standard. Paper chromatographic determination was performed by double ascending development on Whatman no. 1 filter paper in ethylacetate-pyridine-water (10:4:3, v/v). Reducing sugar was detected with aniline hydrogen phthalate reagent (17) and polyalcohol by KI04-p, p'-tetramethyldiaminodiphenylmethane reagent (23). Spots of xylitol, D-arabitol, and ribitol were distinctly separated on the paper showing RF = 1.14, 1.21, and 1.27, respectively, with glucose as standard. Preparation of D-xylulose. A 1-ml amount of a 2-day culture of Acetobacter suboxydans (ATCC 621) grown in a medium described below was inoculated into 500-ml shake flasks each containing 50 ml of the medium composed of 5% D-arabitol, 0.5% polypeptone, 0.1% Yeast Extract (Difco), and 0.1% KH2PO4 (ph 6.0). The flasks were shaken on a reciprocal shaker (140 rev/min, a 7.5-cm stroke) at 30 C. After 2 days of incubation, D-arabitol was almost completely consumed and converted to D-xylulose (93% yield). The clarified broth treated by the method of Neish (9) was concentrated in vacuo at 45 C to a syrup which was extracted with hot 99% ethyl alcohol. After evaporation of ethyl alcohol under vacuum, the resulting syrup was dissolved in water and decolorized by passage through a column of activated carbon. The D-xylulose fraction thus obtained was chromatographically pure, [Ca]20-33.7 (c = 8.99 in water), and the product, D-xylulose, was identified as its p-bromophenylhydrazone derivative. The melting point (128.5 to 129.5 C) was identical with the value given [128 to 129 C, (3)]. Elementary analysis gave the following results: Cu1H1504N2Br; calculated: C, 41.39; H, 4.74; N, 8.78; Br, 25.04; found: C, 41.29; H, 5.01; N, 8.83; Br, 25.20. D-Arabitol was prepared by aerobic fermentation of glucose by yeasts (11). Isolation of xylitol and D-arabitol. After clarification by the method of Neish (9), the cleared broth was concentrated in vacuo at 50 C to dryness. The dried material was extracted with hot 99% ethyl alcohol. When necessary, the extract was treated with Amberlite IRA-400 in the hydroxyl form (19) to remove the remaining reducing sugar. The ethyl alcohol extract gave a crystalline product, and pure pentitol was obtained by recrystallization. RESULTS Screening for yeasts producing xylitol from glucose. A total of 128 yeast strains were screened for their xylitol-producing ability from glucose. Arabitol was the only pentitol produced, but neither xylitol nor ribitol was found. Screening for yeasts producing xylitol from D-xylulose. D-Arabitol has been reported to be produced from glucose in 40 to 50% yields by yeasts (10, 20-22) and to be converted almost quantitatively to D-xylulose by Acetobacter (8, 18). If a yeast grows well on D-xylulose and produces xylitol in good yields, it could be expected that xylitol would be obtained efficiently from glucose through the following sequential fermentation process: glucose -* D-arabitol D-xylulose -* xylitol. Accordingly, production of pentitol from D-xylulose by yeasts was examined. The yeast strains which utilized D-xylulose fairly well and produced good yields of pentitols are shown in Table 1. On the basis of pentitol produced, yeast strains were divided into three groups. Group 1 strains produce xylitol [a representative strain, Candida guilliermondii var. soya (ATCC 20216)], group Il produces D-arabitol [a representative strain, Debaryomyces hansenii (ATCC 20212)], and group III produces xylitol and D-arabitol (a representative strain, Candida guilliermondii 3529). The products were identified from melting points and infrared spectra after crystallization from the broth of representative strains of each group. Effect of cultural conditions on xylitol production. Of group I, Candida guilliermondii var. soya (ATCC 20216) showed the most rapid fermentation with yields of 25.0% of the D-xylulose consumed. The concentration of 6.1 % D-xylulose gave the highest xylitol production (34.2 %, based upon the sugar used). Of several nitrogen sources tested, corn steep liquor was most effective (Table 2). It has been previously reported that the C:N ratio of medium significantly affects polyalcohol production (10, 16, 20). The polyalcohol production by Pichia miso was markedly stimulated in a medium containing 0.1% yeast extract, whereas ethyl alcohol became the principal product when the yeast was cultivated in a medium of 4% yeast extract (16). When the effect of corn steep liquor and Casamino Acids on xylitol production was examined, it was found that 4 to 8% of corn steep liquor markedly increased the xylitol production. The highest yield was 55% of the sugar used. High concentrations of Casamino Acids (1 to 2%) were also effective (Table 3). It has been shown that an excess of inorganic phosphate has a detrimental effect upon the polyalcohol yield in some yeasts (10, 20). The effect of phosphate concentration on xylitol yield was examined in the modified medium B in which KH2PO4 content was varied from 0 to 2%, but the xylitol yields were not affected by any variation of level of the phosphate concentration. Sugar utilization was slower and the xylitol yield was lower at 25 C (5.6% yield), but the yields of xylitol at 30 and 37 C were similar (29.9% yield). Within the range of

Voi.. 18, 1969 XYLITOL FROM GLUCOSE 1033 ph 3.3 to 7.5, the initial ph 6.3 gave the best application of three fermentation processes withyields of xylitol (26.0%). out isolation and purification of the intermediate Production of xylitol from glucose by a sequen- products (D-arabitol and D-xylulose), which tial fermentation process. Our results suggested would make this technique very attractive comthat xylitol would be produced from glucose in mercially. Debaryomyces hansenii ATCC 20212 approximately 15 to 20% yield by a sequential was employed for the first step (glucose> TABLE 1. Survey of pentitol productivity from D-xylulose by yeasts D-Xylulose a Pentitol yield Strain Fermenta- Pentitol based on tion time produced D-xylulose Initial Final used days %~~~~ days I. Yeasts which produce xylitol Saccharomyces rouxii N28 5 6.4 5.5 0.3 33.3 S. rouxii E7 5 6.4 3.6 0.8 28.5 S. rouxii 3281 5 6.4 2.3 1.1 26.8 S. rouxii 3292 5 6.4 2.1 0.1 2.3 S. rouxii 3215 7 6.4 3.0 0.4 11.7 S. rouxii 3217 5 6.4 5.8 0.2 33.3 S. rouxii 3219 6 6.4 4.4 0.9 45.0 S. rouxii 3221 7 6.4 3.9 0.7 28.0 S. rouxii 3222 5 6.4 4.0 1.2 50.0 S. rouxii 3224 5 6.4 5.6 0.2 25.0 S. rouxii var. halomembranis A31 5 6.4 2.5 0.5 12.8 S. acidifaciens S9 5 6.4 1.2 1.4 26.9 S. acidifaciens var. halomembranis H3 5 6.4 2.8 0.6 16.6 S. mellis 3220 7 6.4 5.1 0.6 46.1 Debaryomyces hansenii (ATCC 20220) 4 5.9 0.2 1.5 26.3 Pichia farinosa (ATCC 20210) 3 5.9 2.3 0.6 16.6 P. farinosa (ATCC 20218) 6 5.0 0.1 0.8 16.3 II. Hansenula anomala 4 5.9 0.2 0.3 5.2 H. suaveolens 3 5.6 1.8 0.5 13.2 Endomycopsis chodatii 3 5.9 1.6 0.3 7.0 Candida tropicalis (ATCC 20215) 4 5.0 0.3 0.8 17.0 C. tropicalis 3540 3 4.7 0.4 1.1 25.5 C. guilliermondii var. soya (ATCC 20216) 3 5.0 1.0 1.0 25.0 C. melibiosii (IFO 961) 3 6.4 1.9 0.7 15.5 Candida sp. 3547 3 5.0 0.3 0.7 14.8 Candida sp. 3548 4 5.0 0.3 0.7 14.8 Cryptococcus neoformans 4 5.9 0 0.2 3.3 Yeasts which produce D-arabitol Saccharomyces rouxii 3218 5 6.4 5.1 0.2 15.3 Debaryomyces hansenii (ATCC 20212) 5 5.9 0 2.8 47.4 D. hansenii3170 4 5.9 1.5 0.8 18.2 D. hansenli 3176 5 5.9 0.6 2.9 54.7 D.hansenii 3114 4 5.9 0 2.9 49.1 D.hansenii 3179 5.9 1.9 0.5 12.5 Hansenula subpelliculosa 4 5.9 0 0.9 15.2 Torulopsis famata (ATCC 20214) 4 5.9 1.9 1.5 37.5 T. halophila 5 5.9 3.6 1.5 65.2 T. versatilis (CBS 1752) 6 5.9 3.0 2.3 79.3 Candidapolymorpha (ATCC 20213) 4 5.9 1.1 1.5 31.2 C. parapsilosis (IFO 708) 6.4 1.0 0.8 14.8 C. parapsilosis var. intermedia (IFO 1021) 3 6.4 0.1 1.1 20.7 Trigonopsis variabilis 7 5.0 0.1 1.9 38.7 III. Yeasts which produce xylitol and D-arabitol Candida guilliermondii 3529 6 5.9 2.4 1.8 51.4 a Expressed as grams per 100 ml.

1034 ONISHI AND SUZUKI APPL. MICROBIOL. TABLE 2. Effect of nitrogen sources- Xy~litol Fer- DXloeb yield men D-Xy]ulose based Nitrogen source ta- oln tion D~~~-xylu- lose tion Initial Final sumned hr % Casamino Acids, 0.4%... 69 3.6 1.2 20.6 Ammonium lactate, 0.4%... 60 3.6 0.4 35.3 Urea, 0.05%... 69 3.6 1.4 17.3 Corn steep liquor, 1.6%.... 57 3.6 0.3 46.1 "Medium: 4% D-xylulose, 0.1% KH2PO4, 0.05% Mg54O 7H20, 0.01 % CaC2-2H20, 0.01 % NaCI, and vitamins (biotin, 0.4 pg; calcium pantothenate, 80 pg; inositol, 400 pug; niacin, 80 jug; p-aminobenzoic acid, 40 sg; pyridoxine hydrochloride, 80 pg; thiamine hydrochloride, 80 pxg; and riboflavine, 40 p&g, per 100 ml of medium). The media were adjusted to ph 6.0. Total nitrogen content of corn steep liquor used was 4.08%. TABLE 3. Effect of concentration of Casamino Acids and corn steep liquor upon yield of xylitola Nitrogen source added to the basal medium D-XYluloseb Xylitol yield based on Initial Final D-xylulose consumed Casamino Acids, 0.1%, 5.4 3.3 14.7 Casamino Acids, 0.4%, 5.4 1.4 31.2 Casamino Acids, 1.0%, 5.4 0.3 41.3 Casamino Acids, 2.0%, 5.4 0.3 36.3 Corn steep liquor, 0.4% 5.3 3.1 21.1 Corn steep liquor, 1.6% 5.3 0.9 36.1 Corn steep liquor, 4.0% 5.3 0.2 50.3 Corn steep liquor, 8.0% 5.3 0.2 55.4 a Standard medium B without Casamino Acids and yeast extract was employed as the basal medium. The media were adjusted to ph 6.0. Fermentation time: 69 hr. D-arabitol). A modified medium A was used containing 15% glucose and 4% corn steep liquor. After 4 days of incubation, 13.8 g of glucose was consumed per 100 ml of the medium, and 5.9 g of D-arabitol was accumulated per 100 ml of the broth. The yield of D-arabitol was 42.9% of the glucose consumed. When glucose was almost completely exhausted and yielded D-arabitol, the fermented broth was adjusted to ph 6.0 with NaOH without removing yeast cells and autoclaved at 120 C for 15 min. The broth was thus enriched with the nutrients favorable to the TABLE 4. Production of xylitol from glucose by a sequential fermentation processa Yield of the Amount of product Yield of the prod- Fermentation step sugars or on polyols in substrate the t on- inition thc brothb consumed glnuctale at eachglcs step %; % Glucose 15.5 I. Debaryomyces (77.5 g/ hansenii fermen- 500 ml) tation for 99 hr D-Arabitol 5.3 34.2 34.2 II. Acetobacter suboxydans fermen- 1, tation for 48 hr D-Xylulose 5.0 94.3 32.3 III. Candida guilliermondii var. soya fermentation for 64 hr Xylitol 1.8 41.8 11.6 (9.0 g/ 500 ml) Residual D-xylulose 0.7 a In all fermentation steps, 10 shake flasks, each containing 50 ml of medium (total 500 ml), were run at the same time. At the end of each step, the broth of 10 flasks was mixed, supplemented with corn steep liquor when necessary, adjusted to ph 6.0, and made up to 500 ml. Glucose and D- arabitol at the first and the second steps were exhausted during the respective fermentations. growth of Acetobacter suboxydans, which oxidized the D-arabitol almost quantitatively to D-xylulose in 2 days at 30 C in shake flasks. For the reduction of D-xylulose to xylitol by C. guilliermondii var. soya (ATCC 20216), further addition of 4% corn steep liquor to the broth fermented by A. suboxydans was necessary for the best yield. The fortified broth was adjusted to ph 6.0 with NaOH and autoclaved at 10 lb of steam pressure for 5 min to prevent browning reaction of the medium as much as possible. In 54 hr, 3.6 g of D-xylulose was consumed per 100 ml of the broth and 1.5 g of xylitol was produced per 100 ml of the broth (41.1% yield). Thus, a practical fermentation in shake flasks was run in sequential three-step combination, with optimal conditions for each step. A 9.0-g amount of xylitol was produced from 77.5 g of glucose with a final yield of 11.6% (Table 4). Xylitol was isolated in pure crystalline form from the broth and identified by melting point, elementary analysis, and infrared spectrum.

VOL. 18, 1969 XYLITOL FROM GLUCOSE 1035 DISCUSSION The data presented in this paper demonstrate the practicability of xylitol production from glucose by means of three sequential steps (glucose -- D-arabitol -+ D-xylulose xylitol). -- The xylitol formation was carried out without isolation and purification of the intermediates and yielded 11% xylitol from glucose. The type of pentitol produced from D-xylulose depended on the yeasts which were used for such a reduction. As a result, xylitol, D-arabitol, or a mixture of both were formed. C. guilliermondii var. soya (ATCC 20216), a group I strain (Table 1), produced xylitol (55% yields on the sugar used); D. hansenii (ATCC 20212), a group II strain, produced D-arabitol (50% yields); and Candida guilliennondii 3529, a group III strain, produced almost equal amounts of xylitol and D-arabitol (50% yields of both pentitols). Although xylitol was not found in earlier experiments to be formed from glucose by yeasts (10, 11, 20-22), D-arabitol was the only pentitol detected and identified by us as the dissimilation product of yeasts. D. hansendi (ATCC 20212) and S. rouxii E7, both of which are typical D-arabitol producers, elaborated different pentitols, D-arabitol or xylitol, respectively, from D-xylulose. It has been suggested (1, 4) that D-arabitol formation from glucose by S. rouxii proceeds through two different intermediates, namely, D-xylulose and D-ribulose. This system offers a suitable model for the study of the complex formation of pentitol, especially since it is believed (20) that the intrinsic factors of such formation will most likely be found in the metabolic controls in the cells. ACKNOWLEDGMENTS We thank K. Arima and Y. Ikeda of the University of Tokyo for their guidance. We also thank M. Mogi and N. Iguchi of our Institute for their encouragement. The technical assistance of K. Kouchi is gratefully acknowledged. LITERATURE CITED 1. Blakley, E. R., and J. F. T. Spencer. 1962. Studies on the formation of D-arabitol by osmophilic yeasts. Can. J. Biochem. Physiol. 40:1737-1748. 2. Dische, Z., and E. Borenfreund. 1951. A new spectrophotometric method for the detection and determination of keto sugars and trioses. J. Biol. Chem. 192:583-587. 3. Hough, L., and R. S. Theobald. 1962. D-threo-Pentulose from D-xylose by isomerization and from D-arabinitol by Acetobacter xylinum, p. 94-98. In R. L. Whistler and M. L. Wolfrom (ed.), Methods in carbohydrate chemistry, vol. 1. Analysis and preparation of sugars. Academic Press Inc., New York. 4. Ingram, J. M., and W. A. Wood. 1965. Enzymatic basis for D-arabitol production by Saccharomyces rouxii. J. Bacteriol. 89:1186-1194. 5. Lang, K. 1964. Die ernrhrungsphysiologischen Eigenschaften von Xylit. Int. Z. Vitaminforsch. 34:117-122. 6. Lohman, R. L. 1957. The polyols, p. 245-246. In W. Pigman, ed., The carbohydrates: chemistry, biochemistry and physiology. Academic Press Inc., New York. 7. Mellinghoff, C. H. 1961. Uber die Verwendbarkeit des Xylit als Ersatzzucker bei Diabetikern. Klin. Wochenschr. 39: 447. 8. Moses, V., and R. J. Ferrier. 1962. The biochemical preparation of D-xylulose and L-ribulose. Details of the action of Acetobacter suboxydans on D-arabitol, ribitol, and other polyhydroxy compounds. Biochem. J. 83:8-14. 9. Neish, A. C. 1952. Analytical methods for bacterial fermentations, 2nd revision. National Research Council of Canada. N.R.C. Bull. 2952. 10. Nickerson, W. J., and R. G. Brown. 1965. Uses and products of yeast and yeast-like fungi. Advan. Appl. Microbiol. 7.225-272. 11. Onishi, H. 1960. Studies on osmophilic yeasts. VIH. Polyalcohol production by various genera and species of yeasts. Bull. Agr. Chem. Soc. Japan 24:131-140. 12. Onishi, H., and M. B. Perry. 1965. The production of mesoglyceroido-heptitol and D-glycero-D-ido-heptitol by Pichia miso. Can. J. Microbiol. 11:929-934. 13. Onishi, H., and T. Suzuki. 1966. The production of xylitol, L-arabinitol and ribitol by yeasts. Agr. Biol. Chem. 30:1139-1144. 14. Onishi, H., and T. Suzuki. 1968. Formation of dulcitol in aerobic dissimilation of D-galactose by yeasts. J. Bacteriol. 95:1745-1749. 15. Onishi, H., and T. Suzuki. 1968. Production of D-mannitol and glycerol by yeasts. Appl. Microbiol. 16:1847-1852. 16. Onishi, H., N. Saito, and I. Koshiyama. 1961. Studies on osmophilic yeasts. XI. Various factors affecting polyalcohol production by Pichia miso. Agr. Biol. Chem. 25:124-130. 17. Partridge, S. M. 1949. Aniline hydrogen phthalate as a spraying reagent for chromatography of sugars. Nature (London) 164:443. 18. Prince, R., and T. Reichstein. 1937. Synthesis of d-xylosonic acid. Helv. Chim. Acta 20:101-109. 19. Roseman, S., R. H. Abeles, and A. Dorfman. 1952. Behavior of carbohydrates toward strongly basic ion-exchange resins. Arch. Biochem. Biophys. 36:232-233. 20. Spencer, J. F. T. 1968. Production of polyhydric alcohols by yeasts. Prog. Ind. Microbiol. 7:1-42. 21. Spencer, J. F. T., and H. R. Sallans. 1956. Production of polyhydric alcohols by osmophilic yeasts. Can. J. Microbiol. 2:72-79. 22. Spencer, J. F. T., J. M. Roxburgh, and H. R. Sallans. 1957. Factors influencing the production of polyhydric alcohols by osmophilic yeasts. J. Agr. Food Chem. 5:64-70. 23. Yoda, A. 1952. A new method for detection of non-reducing sugars and polyalcohols on paper chromatogram. J. Chem. Soc. Japan. 73:18-19.